Agent skills for Zed

Reduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT.

Guide for maintaining the MassGen model and backend registry.

Allocate models for one task from the runtime catalogue, with operator confirmation, light-worker eligibility rules, and a…

Generate a reproducible, runnable PyTorch training repo for a medical-imaging task — segmentation, classification, detection…

When to activate: model serving, inference, FastAPI ML, BentoML, Triton, vLLM, model deployment, online inference, batching

Vet the concrete third-party model a study will be built on — this repository, this revision, this checkpoint — not the architecture family.

You are an expert Rails model test developer.

Fires before ANY agent or workflow dispatch that names a model or effort — before a `RuntimeAssignment` is written — routing workflow…

Design or audit the clinical-validation study for an engineer-built medical-imaging model (segmentation, classification, or detection)…

Load when handling moderation issues, conflict resolution, rule enforcement, or developing moderation policies.

Use when modern C++ features from C++11/14/17/20 including auto, lambdas, range-based loops, structured bindings, and concepts.

When to activate: C++, modern C++23, RAII, smart pointers, move semantics, concepts, ranges, structured bindings, value semantics

Use when writing, reviewing, or modernizing Kotlin code with recent language versions (2.0–2.4.x, including 2.4.10 and 2.4.20-Beta1)…

Write modern PHP 8.x code — typed properties, enums, readonly classes, match expressions, named arguments, union/intersection types…

Enforces high-performance Python 3.10+ standards using functional paradigms, strict typing, and high-speed serialization.

Generates a clean, professional single-page A4 resume optimized for print and PDF export.

Build modern Shiny dashboards and applications using bslib (Bootstrap 5).

Run educational code reviews against your project's quality standards.

Audit and maintain documentation coverage across any codebase using the three-layer system (README.md for humans, AGENTS.md for AI agents…

Track meaningful corpus changes using frontmatter, doc/_meta/generation-log.md and, only for major corpus-level changes…

Use when editing an existing skill's SKILL.md — adding sections, updating descriptions, restructuring, fixing gaps, or closing loopholes…